通过因果学习减少文本分类中的虚假关联，提升模型鲁棒性。
Fighting Spurious Correlations in Text Classification via a Causal Learning Perspective
- 基于反事实推理选择因果特征，抑制无关特征干扰
- 在无群体标签场景下达到当前最优性能，部分媲美有标签方法
- 适合需要高泛化能力的文本分类任务，尤其应对分布外数据
在文本分类任务中，模型常依赖虚假相关性进行预测，错误地将无关特征与目标标签关联，这限制了模型的鲁棒性和泛化能力，尤其在分布外数据上表现不佳。为此，我们提出因果校准鲁棒分类器（CCR），旨在减少模型对虚假相关性的依赖，提升鲁棒性。方法结合基于反事实推理的因果特征选择与无偏逆倾向权重（IPW）损失函数。通过聚焦于因果特征的选择，确保模型在预测时较少依赖虚假特征。我们从理论上证明了该方法的有效性，并实证表明，CCR在无群体标签的方法中达到最先进性能，某些情况下甚至可媲美使用群体标签的模型。
原文摘要 · Abstract (English)
In text classification tasks, models often rely on spurious correlations for predictions, incorrectly associating irrelevant features with the target labels. This issue limits the robustness and generalization of models, especially when faced with out-of-distribution data where such spurious correlations no longer hold. To address this challenge, we propose the Causally Calibrated Robust Classifier (CCR), which aims to reduce models' reliance on spurious correlations and improve model robustness. Our approach integrates a causal feature selection method based on counterfactual reasoning, along with an unbiased inverse propensity weighting (IPW) loss function. By focusing on selecting causal features, we ensure that the model relies less on spurious features during prediction. We theoretically justify our approach and empirically show that CCR achieves state-of-the-art performance among methods without group labels, and in some cases, it can compete with the models that utilize group labels.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。